Norms and Attitudes About Being an Active Bystander: Support for Telling Adults About Seeing Knives or Guns at School Among Greater London Youth

Abstract

A detailed understanding of the factors associated with support among youth for reporting a knife or gun at school to an adult is essential to inform violence prevention initiatives. However, no studies have empirically assessed attitudes about support for reporting among secondary school students in Greater London nor perceived norms about such support among peers. Thus, this study explores whether students misperceive peer norms about support for telling adults about seeing weapons at school. Anonymous surveys were completed by 7,401 youth (52% female; 43% White; mean age 11.8 years) in school years 4 to 11 in 45 school cohorts in a greater London borough between 2007 and 2012. Students reported both personal support about reporting weapons to several categories of adults and whether they perceived most other students at their school to support reporting weapons to adults in each category. Most students (64% to 78% on average) in most cohorts personally thought that students should report seeing a weapon at school to a head teacher, police officer, teacher or counsellor, and parent or another adult relative. However, 34% to 44% of students erroneously thought that the majority of their peers did not support reporting to these adults. Perceived norms predicted personal support for reporting, adjusting for the prevalence of actual support at one’s school and other factors. Pervasive norm misperceptions about reporting may contribute to a less safe environment.

INTRODUCTION

Violence perpetration by youth and among youth is a serious problem. Violent physical fighting and weapon-carrying are associated with immediate and long-term physical and emotional injury and negative health outcomes (Walsh et al. 2013). In addition, fear of violence is associated with decreased academic performance (Barrett, Jennings, and Lynch 2012). A recent study in the United Kingdom found that children and adolescents under 16 years were more likely to be stabbing victims than people aged 16 to 25 years, and that they were more likely to be stabbed on school days, in the hours immediately after school, and within 5km of school (Vulliamy et al. 2018).

A review of prevented school killings from 1900 to 2016 found that most were prevented because students reported the threat to school or law enforcement authorities (Stallings and Hall 2019). Unfortunately, however, students often do not report potential threats to an adult or someone in a position of authority even though students often have knowledge about the presence of a potentially violent threat at school before school authorities are aware of a threat (Vossekuil, Fein, Reddy, Borum, and Modzeleski 2002; Madfis 2014). As personal attitudes about an action are strong predictors of intentions to act and subsequent personal behaviors (Ajzen 2001, 2005), a detailed understanding of the sociodemographic and potential protective factors associated with support among youth for reporting a knife or gun at school to an adult is essential to inform violence prevention initiatives. No studies have empirically assessed such attitudes among secondary school students in Greater London. However, news analyses have highlighted increased weapons-related violence in this context (Torjesen 2018) and editorials have called for greater attention to reducing weapons-related violence among youth (Middleton and Shepherd 2018). Thus, this study assesses support for being an active bystander among adolescents in Greater London with a focus on social norms as critical factors for building personal support for reporting.

Recent data from the United Kingdom indicates that over a 12 month period, 6% of 10–15 year old youth knew someone who carried a knife for protection (Allen and Audickas 2018). Upon seeing or hearing about a knife or a gun at school, students would ideally speak up and tell an adult when they have knowledge about a weapon at school. Decades of studies have shown that adolescents are strongly influenced by their peers and that this influence increases from middle school through high school as youth spend more time with peers (Brechwald and Prinstein 2011). People inherently look to others for approval and to know what to believe and approve of as well as how to act especially in contexts of uncertainty (Cialdini and Trost 1998). The attitudes about reporting that youth perceive to be typical among peers within their school setting may contribute to personally having an attitude that approves of students telling a school official, police, or other adult about seeing a weapon such as a knife or gun at school.

Social Norms and Norm Misperception: A Conceptual Framework

The Theory of Reasoned Action and the Theory of Planned Behavior propose that subjective norms, what an individual thinks others expect someone to do or believe, are important antecedents to intentions and behaviors and that they are associated with attitudes about a given behavior (Montano and Kasprzyk 2015). Relatedly, perceived injunctive norms are what individuals perceive to be the most common attitudes about what is acceptable and right (that is, beliefs about what people ought to do) held by a majority of a group (Cialdini, Reno, and Kallgren 1990). Social Norms Theory applies a sociological as well as psychological lens to the role of social norms in predicting personal attitudes and behavior (H. W. Perkins 2003a). It distinguishes between actual norms of the group at a macro level and perceived norms at a micro or subjective level (what is the empirically most typical behavior within a reference group versus individuals’ perceptions about what is the most typical behavior among a reference group). This theory predicts that although what a person perceives to be normative among their peers may not actually be the normative attitude or behavior, these erroneous perceptions may still have great influence on personal attitudes and behaviors in spite of their inaccuracy (H. W. Perkins 2014). Substantial empirical work based on these theories has found that perceived peer norms are associated with personal attitudes and behaviors, for such topics as adolescent and college student alcohol and tobacco use in the United States (H. W. Perkins 2003b), cannabis use among European university students (Dempsey, McAlaney, and Bewick 2018), non-prescription drug use among European university students (Lehne et al. 2018), unintended pregnancy among young mothers (Compernolle 2017), junk food consumption and sedentary habits among teens (Rice and Klein 2019), delinquency among rural youth in the United States (Cotter and Smokowski 2016) and among youth in the Netherlands (Young and Weerman 2013), intimate partner violence acceptance and propensity (Mulla et al. 2018), and gun carrying among youth in the northeastern United States (Hemenway, Vriniotis, Johnson, Miller, and Azrael 2011). Moreover, students often overestimate the prevalence and strength of unhealthy and risky attitudes and behavior among peers and perceive these attitudes and behaviors to be the norm even though healthy and protective attitudes and behaviors are most often actually normative in their peer groups (H. W. Perkins and Perkins 2018).

Furthermore, separate measurement of these social norm constructs allow for assessment of the extent to which perceived norms predict attitudes and behavior above and beyond the predictive contribution of actual norms. Some multivariate studies have found that perceived norms are even more strongly associated with personal attitudes and behaviors than are actual peer norms for such topics as sugar sweetened beverage consumption and fruit and vegetable intake among middle and high school students across schools in the United States (J.M. Perkins, Perkins, and Craig 2018), adolescent body weight across secondary school cohorts in the United Kingdom (J.M. Perkins, Perkins, and Craig 2010) and in the United States (J.M. Perkins, Perkins, and Craig 2015), alcohol use across schools in the United States (H. W. Perkins, Haines, and Rice 2005) and among emerging adults in the United States (Simons-Morton, Haynie, Bible, and Liu 2018), and across institutions of higher education in Canada (H. W. Perkins 2007), and road traffic risk behaviors among young German drivers (Geber, Baumann, Czerwinski, and Klimmt 2019).

If both substantial injunctive norm misperception and an association between the perceived norm and personal attitude are present, then there may be an opportunity to change attitudes, and ultimately behavior, by correcting students’ misperceptions about peer norms. The Social Norms Approach to intervention emphasizes designing messages about existing positive norms among a peer group based on real data as opposed to focusing on risky behavior or bad outcomes (H. W. Perkins and Perkins 2018). Doing the latter could make that unhealthy behavior or attitude seem more common than is true and inadvertently increase the harmful behavior/attitude. Instead, communicating actual positive norms via various methods to a targeted population may support an overall more healthy/positive culture and may reduce any misperceived stigma about the positive behavior/attitude that may exist (e.g., “snitching”).

Over the past several decades, many studies have employed a social norms approach to assessing the impact of misperceived norms about risky alcohol consumption on personal drinking using designs that allow for causal assumptions. They have found that introducing information about actual norms has led to changes in perceived norms, which has led to personal changes in alcohol consumption and attitudes. For example, the causal impact of perceived norms on behavior has been demonstrated in a longitudinal study of a media campaign among college students to correct misperceptions and reduce high risk drinking (Turner, Perkins, and Bauerle 2008), in experiemental design interventions providing personalized normative feedback to reduce alcohol consumption among randomly assigned college students (Neighbors, Larimer, and Lewis 2004; Neighbors et al. 2010) and young veterans (Pedersen, Parast, Marshall, Schell, and Neighbors 2017), and in an experiment randomly assigning universities to a social norms media campaign intervention or to a control condition to reduce risky drinking (DeJong et al. 2006). In addition, a state-wide study using a quasi-experimental design found that a social norms media campaign to correct norm misperceptions produced a reduction in drinking and driving among young adults over an 18 month period when comparing intervention and control counties (H. W. Perkins, Linkenbach, Lewis, and Neighbors 2010). Studies on other topics such as gambling (Larimer et al. 2012; Neighbors et al. 2015), sexual risk taking (Lewis et al. 2014; Lewis, Rhew, Fairlie, Swanson, Anderson, and Kaysen 2018), smoking initiation (Linkenbach and Perkins 2003), bullying (Paluck, Shepherd, and Aronow 2016), sun prevention (Reid and Aiken 2013), and reporting corruption (Blair, Littman, and Paluck 2019) also support changing perceived norms to increase positive behavior.

Bystander Support and Social Norms

Based on this body of work, it is plausible that a student might perceive that most peers at school do not approve of reporting a weapon at school to an adult even when support for reporting is normative in that context. This misperception may then contribute to personally having an attitude against telling a school official, police officer, or another adult about seeing a weapon such as a knife or gun at school. Alternatively, if most students perceive other peers to support being an active bystander, then they may be more likely to personally support reporting. Thus, there are three critical questions to answer for the development of violence prevention interventions focusing on active bystander support and action. First, how common are actual positive norms about reporting weapons to adults among youth, especially in contexts where knife or gun use, bullying, and other forms of violence are of concern to school officials. Second, to what extent are such positive norms misperceived by students in general and by specific sub-groups of youth. Third, are norm perceptions about support for reporting (i.e., perceived injunctive norms) associated with personal support for reporting while accounting for any association between actual local norms for reporting and personal support.

Some research on support for and engagement in active bystander intervention and reporting among youth and young adults in the United States offers evidence that perceived social norms about reporting hurtful circumstances or potentially violent threats are underestimated and that perceived norms are associated with personal attitudes towards being an active bystander. For example, one study about bullying and victimization found that youth underestimated support for intervening on bullying among peers and those who did so were less likely to report intervening (Sandstrom, Makover, and Bartini 2013). Another study found an association over time between perceived peer injunctive norms about defending victimized peers and personally defending victimized classmates (Kollerová, Yanagida, Mazzone, Soukup, and Strohmeier 2018). Additionally, youth have underestimated willingness of peers to challenge aggression in hypothetical scenarios (Mulvey and Killen 2016) as well as underestimated support among peers for reporting bullying-related harassment to adults (H. W. Perkins, Craig, and Perkins 2011). Intervention studies on bystander support and action related to sexual assault have found that reducing misperceptions about active bystander support and action (seen as more rare than in reality) have been associated with evidence of improved personal attitudes about bystander engagement and subsequent intentions and actions among college students (Zapp, Buelow, Soutiea, Berkowitz, and DeJong 2018; Mennicke, Kennedy, Gromer, and Klem-O’Connor 2018). Finally, one U.S.-based study of youth across six states specifically looked at reporting weapons-related threats and found that many middle and high school students think that their peers do not approve of reporting a weapon at school to someone with authority despite most students indicating support for students to do so (J.M. Perkins, Perkins, and Craig 2017). However, no studies have empirically assessed actual and perceived peer injunctive norms about reporting weapons to several categories of adults across a diverse range of students in the United Kingdom.

The Current Study

This study addressed two empirical gaps in the literature. First, the prevalence of misperceived norms about support for being an active bystander was assessed among students in school years 4 to 11 across different sociodemographic factors, social activity groups, and a diversity of school contexts in Greater London. Specifically, perceived and actual injunctive norms about peer attitudes toward telling different adults at school or a parent/other adult relative about seeing a weapon such as a knife or a gun at school were investigated and compared. Given the tendency in prior research for students to erroneously believe that positive or healthy attitudes and behavior are not the norm, it was hypothesized that the actual injunctive norms found in this study population would support reporting to each of the adults in most, if not all, school contexts, but that many students would perceive their local school peers to exhibit a negative injunctive norm; that is, many students would misperceive their peer norm by erroneously thinking most peers do not support reporting. Second, the association between the perceived injunctive norm about reporting to a specific adult and personal support for reporting to that adult was assessed while adjusting for the strength of the actual injunctive norm and other factors. It was expected that an individual’s perception of the peer injunctive norm would more strongly predict personal attitude toward reporting a weapon to different adults than the strength of the actual injunctive norm.

METHODS

Procedure

Both upper level and primary schools in a large borough of Greater London were contacted by the local borough’s educational personnel in charge of promoting student well-being. They were invited to participate in an assessment of student-related concerns through an online survey made available for use at each school in 2007, 2010, 2011 and 2012. As several schools participated in the survey in multiple years, each survey administration represented a single school cohort. Upper level schools most often surveyed students in all years present in the school while most primary schools surveyed students in school year 5 and above. A few schools included students in year 4 when general reading comprehension was adequate to do so. Local schools obtained parent/guardian consent for student participation via their standard procedures.

In group settings in school where a computer was available for each student, students were provided with general information about the online survey and informed that it was voluntary and either anonymous or confidential. In one administration format, every student in a specific group session was publicly provided with the same url address and password to access the survey in order to assure students of their anonymity in taking the survey. The password was then changed between sessions so that no student could access the survey and submit additional responses after leaving the survey session. In the other administration format used in later years, the survey was conducted by guaranteeing student confidentiality and by providing individual passwords. Passwords assigned to students could only be used once and were given to them as they arrived to take the survey in the computer room. Confidentiality was assured by the passwords being separated from respondent results as responses were submitted by students. A monitor was present to provide initial instructions about survey access and make sure that students did not interact while taking the survey.

Sample

As some schools participated in the survey multiple times, only data from school cohorts where no year levels overlap in survey administrations were used. In addition, the cohort response rate to the survey and specifically to the section of questions that were the main subject of this study (attitudes and perceived norms about reporting weapons to adults) had to be above 50% because this study was focused on misperceptions about actual injunctive norms (the attitudes that the majority of individuals in a cohort personally reported). Thus, the sample included 7,934 students across 45 cohorts representing 35 schools with an average survey participation rate of 87% within a cohort (standard deviation (s.d.) = 10 percentage points and a minimum response rate of 63%). Within the 45 included cohorts, there were 533 students (6.7%) who did not respond to the section of questions about attitudes and perceived norms about reporting weapons. (On average, there was 6% non-section respondents across the cohorts with an s.d. of 4.3 percentage points). Thus, the overall average question response rate was 82% (s.d. = 11 percentage points and minimum = 57%). (This latter response rate was calculated as the number of students within a cohort that participated in the survey and also responded to the section of questions used in this study divided by the total number of students in a cohort targeted for participation.) The students who did participate, but who did not provide data about reporting weapons, were distributed across demographic categories. For example, 8% were male and 4% were female and 5.7% were in primary schools and 7.6% were in secondary schools. Across racial/ethnic groups included in this study, the prevalence of these section non-respondents ranged from 2% to 4.5%. After excluding these non-section respondents, the final analytical sample was 7,401 students.

Respondents for this study came from 25 public schools and 10 other types of schools (of which 4 were not coed). Twenty-seven schools were primary schools enrolling students in school years 6 and lower and 8 schools were secondary schools enrolling students in school years 7 or higher. Although most schools were primary level, they were smaller in size compared to schools with older students. The student body size of the cohorts ranged from 163 to 1940 students (mean = 482; s.d. = 337) with 26 of the 27 primary schools having less then 500 students in the population and all of the upper level schools having 680 students or more.

Measures

The survey initially instructed students to think about students in their specific school when asked questions about most other students.

Personal support for reporting weapons (i.e., personal attitude about reporting).

Students were asked: “Who do you think students should tell if they see a weapon (knife or gun) at school? Students were shown a list of response options that was titled “I think students should …” where the options included tell a head teacher, tell a teacher or counsellor, tell a police officer, tell a parent or another adult relative, or not tell anyone. They were instructed to tick all options that apply. Dichotomous variables were created where personally believing that students should report seeing a weapon to each of these options was coded as 1 and 0 otherwise. The items were judged by the research team to have face validity as direct measures of personal support for reporting weapons to the particular indicated adult category. They were drawn from a previous research study in the United States measuring student attitudinal support for reporting weapons to adults (J.M. Perkins et al. 2017a), and were modified only slightly for spelling and word variations in the British context.

Prevalence of support at school for reporting weapons (i.e., actual extent of peer support for reporting).

Individuals’ responses about each personal support for reporting variables were aggregated, separately, within a cohort to calculate variables representing the percentage of students within a cohort who personally thought students should tell that adult about seeing knives or guns at school. The low end of a 95% confidence interval around each of these prevalence variables per cohort was also calculated to represent the ‘minimum prevalence of support’ for reporting to each of the specific adult categories that would likely be present in the cohort population. This information was used to assess whether any effect of sampling variation would likely impact results.

Actual student norms about support for reporting (i.e., actual injunctive norms about reporting).

Based on the continuous variables representing prevalence of support, dichotomous variables were created to indicate the actual injunctive norms for the cohort. As this study is substantively focused on what the majority of a group supports, the majority was statistically defined as more than 50%. Thus, if more than 50% of students within a cohort supported telling an adult, then the actual injunctive norm variable was coded as 1, otherwise it was coded as 0 if only 50% or less supported reporting to that adult. Previous studies on social norms have measured actual norms in a similar way by calculating what more than 50% of the sample representing a defined population reports believing or doing for dichotomous variables or by using the median for ordinal variable categories (H. W. Perkins and Craig 2012; J. M. Perkins et al. 2017a; J.M. Perkins et al. 2017b; J.M. Perkins et al. 2018). In addition, trichotomous actual injunctive norm variables were created based on a prevalence of 76% to 100% support (highly normative support), 51% to 75% (moderately normative support, and 50% or less support (support not normative). Finally, an additional set of trichotomous variables was created using the ‘minimum prevalence of support variables’ to identify cohorts where support would still be normative (among more than 50% of the cohort) in almost all potential samples (97.5% based on the lower limit of the 95% confidence interval) of a hypothetical repeated sample condition.

Perceptions of student norms about support for reporting weapons (i.e., perceived injunctive norms about reporting).

Immediately after the personal support questions, students were asked: “And what would most other students say?” Students were shown a list of response options titled “Most other students would say …”, which included tell a head teacher, tell a teacher or counsellor, tell a police officer, tell a parent or another adult relative, or not tell anyone. They were instructed to tick all options that apply. This format of asking about perceived norms by asking what most other student peers thought about or did with regard to the topic using the same language as used for items asking about personal attitudes or behavior is based on previous social norms studies assessing students’ perceptions of the most typical behavior and attitudes among a specified peer group. For example, several studies have used this survey format among youth and young adults in different countries including youth in the United States (H. W. Perkins et al. 2011; J.M. Perkins et al. 2017a; J.M. Perkins et al. 2018) and the United Kingdom (J. M. Perkins et al. 2010), and college students in the United States (H. W. Perkins et al. 2005; H. W. Perkins 2007; H. W. Perkins and Craig 2012), Canada (H. W. Perkins 2007), and several European nations (Helmer et al. 2014; McAlaney et al. 2015; Pischke et al. 2015; Dempsey et al. 2016). Dichotomous variables were created where perceiving that most other students should report seeing a weapon to each of these options was coded as 1 and 0 otherwise. Six percent of students did not respond to this question.

Norm misperception (i.e., misperceived injunctive norms about reporting).

Students’ perceptions of peer norms about support for reporting weapons were compared with actual norms among their own cohort. This method for calculating the prevalence of norm misperception is based on previous work among youth (J.M. Perkins et al. 2017a; J.M. Perkins et al. 2018), college students (H. W. Perkins and Craig 2012; H. W. Perkins 2007; H. W. Perkins et al. 2005) and a general adult population (J. M. Perkins et al. 2017b; J. M. Perkins et al. 2019). Given our substantive interest in misperception of positive norms and our hypothesis that most, if not all, cohorts would exhibit majority support for reporting to each of the adults, norm misperception was coded as 1 if students perceived most peers to not support reporting when the actual injunctive norm was to support reporting. The variable was coded as 0 if students perceived support to be normative and the prevalence of support was greater than 50%. In addition, this calculation of norm misperception was repeated using the actual norm variables based on the ‘minimum prevalence of support’ as the comparison to create a more conservative set of norm misperception variables for sensitivity analyses.

Other factors.

Several measures of student participation in extracurricular activities, school experiences, and sociodemographic characteristics were included because an aim of this study was to assess the extent of norm misperception across different sub-groups of students. If substantial norm misperception exists across many kinds of student groups, then a population-wide intervention could be implemented. Thus, data were recorded on participation in student council, school sports, school choir or band, after school club, after school activity in the community, and a part-time job (missing 0.7% of responses for each activity). These kinds of activities provide opportunity for development of relationships among peers, students, teachers, and other adults, which theoretically support favorable attitudes towards reporting to keep these communities safe and productive. In addition, students reported whether they had ever felt bullied during the school year (missing 1.9% of responses), and indicated whether they agreed with the statement, “I feel that teachers care about me” (missing 1.4% of responses). Past studies have found that school climate and feeling connected to teachers and peers at school are associated with breaking a code of silence and reporting threats to adults (Pollack, Modzeleski, and Rooney 2008; Daniels et al. 2009).

Finally, academic self-evaluation, parental education, race/ethnicity, gender, and year in school were also collected. Students rated their academic performance as very poor or poor (combined for this study), average, good, or excellent in response to the question ‘how well do you feel you do in your school work’. An ‘unknown’ category was created to retain the 3.7% of students in analyses who did not respond to the question. Highest parental education was classified into the following categories: low (primary or secondary education only); medium (college / vocational / training education); high (university or more); and, Unknown. The “unknown” category was created to retain the 52% of students who did not receive the parental education question in early versions of the survey or who did not respond to the question at all in later years. The question about race/ethnicity provided choices of 20 detailed categories, subsequently condensed into six groups: White, Asian or Asian British, Black or Black British, Mixed, Other, and Unknown. The “unknown” race category was created to retain the 10% of students who indicated ‘unknown’ or ‘I prefer not to say’ or who did not respond to the race question at all. A response about gender was missing for 1.0% of students and the year in school response was missing for 1.1% of students. In total, there were 326 respondents (4.4%) out of 7,401 who were missing at least one response to factors included in this study (excluding the variables for which ‘unknown’ was a specific category). These students were still included in most analyses because their data on perceived norms and self-reported attitudes could be included for most calculations and analyses about actual and misperceived injunctive norms. However, these respondents were not included in the regression analyses.

Analytic Strategy

First, actual injunctive norms at the local cohort level were reported and the distribution of support for reporting to different adults (i.e., mean prevalence and standard deviation) was calculated across cohorts to describe the extent of support across different contexts. In addition, the prevalence of support for reporting to different adults was calculated within student sociodemographic characteristics and participation in extracurricular activities to indicate whether support was common across different types of students. Statistical testing for differences between sub-categories was conducted (correcting for the inflated chance of a Type I error due to the number of significance tests involved using a Bonferroni correction).

Second, the distribution of norm misperceptions (i.e., mean prevalence and standard deviation) was calculated within cohorts of different characteristics. In addition, the prevalence of norm misperceptions across student characteristics was calculated and statistical testing was conducted to assess for differences between groups (again using a Bonferroni correction). The intent was to show the extent to which norm misperception permeated different contexts and student groups. For these analyses, data from all 7,401 students were used when available. The prevalence of norm misperceptions was also recalculated using the more conservative set of norm misperception variables (which were themselves calculated using the estimate of the actual norm favoring reporting for each category of adult based on the low end of the estimated confidence interval for prevalence of support in each cohort).

To assess the relationship between personal support for reporting to a specific category of adult and peer injunctive norms (both perceived and actual), a multilevel logistic regression model was fit that accounted for clustering of outcomes within cohorts and included variables representing these perceived and actual peer norms. The initial models included whether a student perceived that the majority of students in the cohort were in support of reporting a weapon at school to that category of adults, several other characteristics of students, and survey year. In addition, these models included the prevalence of actual support in one’s cohort for reporting a weapon to a category of adults. The models were then re-fit using the ‘minimum prevalence of support’ variables. However, as the premise in this study about peer norms being impactful is based on whether support is normative, the main regression models included the strength of the actual injunctive norm (instead of the continuous prevalence variable) represented as highly normative, moderately normative, and not normative. All models excluded the 4.4% of students who had at least one missing observation.

RESULTS

Actual Injunctive Norms

A majority of students in all 45 cohorts thought students should report seeing a weapon at school to a head teacher with 78% of students on average across the cohorts (s.d. = 10 percentage points) believing that students should do so (Table 1). Likewise, most students in 44 out of 45 cohorts thought students should report a weapon to a police officer, with 68% supporting this action on average (s.d. = 9 percentage points). In 36 out of 45 cohorts and in 32 out of 45 cohorts, a majority of students (64%) supported telling a teacher or counsellor and a parent or another adult relative, separately, about seeing a weapon at school. Thus, the actual injunctive norms across most cohorts supported reporting a weapon to these adults. When cohorts were stratified by primary vs. secondary school status or the prevalence of actual support within a cohort being highly normative vs. moderately normative, the majority of students in most cohorts also supported reporting a weapon to these adults. In the few cohorts where support was not the actual injunctive norm, the extent of students who supported reporting fell just shy of being normative (45%, on average, supported telling a teacher, and 45%, on average, supported telling a parent or another adult relative).

Table 1.

Prevalence of support for telling adults about seeing a weapon at school across 45 school cohorts in a greater London borough.

Cohort type	Adults to which a student should report seeing a weapon (knife or gun) at school	Cohorts in which the actual injunctive norm supported reporting to a specific category of adults		Prevalence of support for reporting weapons to a specific category of adults in cohorts where the actual injunctive norm supported reporting
		N	(% of total)	Mean prevalence	(standard deviation in percentage points)
All	Head teacher	45	(100%)	78%	(10 p.p)
	Police officer	44	(98%)	68%	(8 p.p)
	Teacher or counsellor	36	(80%)	64%	(7 p.p)
	Parent or another adult relative	32	(71%)	64%	(8 p.p)
Primary level (school years 6 and less)	Head teacher	37	(100%)	81%	(7 p.p)
	Police officer	36	(97%)	70%	(7 p.p)
	Teacher or counsellor	31	(84%)	66%	(7 p.p)
	Parent or another adult relative	31	(84%)	64%	(7 p.p)
Upper level (school years 7 and up)	Head teacher	8	(100%)	62%	(6 p.p)
	Police officer	8	(100%)	57%	(5 p.p)
	Teacher or counsellor	5	(63%)	57%	(4 p.p)
	Parent or another adult relative	1	(13%)	52%	n/a
Prevalence of support for reporting was 76–100% (i.e., highly normative)	Head Teacher	32	(100%)	83%	(5 p.p)
	Police officer	10	(100%)	79%	(3 p.p)
	Teacher or counsellor	3	(100%)	81%	(3 p.p)
	Parent or another adult relative	3	(100%)	80%	(3 p.p)
Prevalence of support for reporting was 51–75% (i.e., moderately normative)	Head teacher	13	(100%)	65%	(6 p.p)
	Police officer	34	(100%)	64%	(7 p.p)
	Teacher or counsellor	33	(100%)	63%	(6 p.p)
	Parent or another adult relative	29	(100%)	62%	(6 p.p)

Notes: The ‘actual injunctive norm supported reporting’ was defined as where more than 50% of a cohort supported reporting to the specific category of adults. There were a few cohorts where support for reporting was not normative. That is, in 1 out of 45 cohorts, only 50% of the cohort supported reporting a weapon to a police officer, in 9 out of 45 cohorts the prevalence of support for reporting a weapon to the teacher or counsellor was no larger than 50% (44% on average), and in 13 out of 45 cohorts no more than 50% (44% on average) supported reporting a weapon to a parent or another adult relative (s.d. for both = 5 percentage points)

At the individual level, 76% of students supported telling a head teacher about seeing a weapon at school, 67% supported reporting it to a police officer, 61% supported reporting it to a teacher or counsellor, and 56% supported reporting it to a parent or another adult relative. Furthermore, most students across sociodemographic groups supported telling each type of adult about a weapon at school though there were some differences (Table 2). For example, more females supported telling a head teacher, teacher or counsellor, and parent or another adult relative, separately, than did males. The prevalence of support for reporting to each of type of adult was greater among the White students that comprised 43% of the overall sample in this study than among students of other race/ethnic identities. And, the prevalence of support for reporting was higher among students who reported having been bullied, feeling that teachers care, and participating in school clubs or student government.

Table 2.

Prevalence of support for telling a specific category of adults about seeing a weapon (knife or gun) at school by socio-demographic characteristics of youth in Greater London.

			Prevalence of personal support for reporting weapons to a specific category of adults
Student characteristics	n	(%)	Head teacher	p-value	Police officer	p-value	Teacher or counsellor	p-value	Parent or another adult relative	p-value
All	7401		76%		67%		61%		56%
Gendera				0.006		0.052		<.001		0.001
Female	3838	(52%)	78%		66%		62%		58%
Male	3487	(47%)	75%		69%		59%		54%
Year in schoola				<.001		<.001		<.001		<.001
Year 4	235	(3%)	86%		67%		66%		64%
Year 5	1716	(23%)	85%		70%		64%		63%
Year 6	1387	(19%)	86%		77%		68%		67%
Year 7	1011	(14%)	78%		69%		60%		57%
Year 8	971	(13%)	71%		66%		56%		50%
Year 9	882	(12%)	66%		59%		53%		47%
Year 10	875	(12%)	60%		57%		56%		41%
Year 11	230	(3%)	57%		57%		53%		37%
Race/Ethnicityb				<.001		<.001		<.001		<.001
White	3152	(43%)	81%		72%		67%		60%
Black	1588	(21%)	69%		61%		54%		51%
Asian	937	(13%)	77%		67%		61%		51%
Other	782	(11%)	70%		66%		55%		53%
Mixed	191	(3%)	70%		67%		60%		55%
Unknown	751	(10%)	77%		65%		55%		56%
Parental educationb				0.047		0.016		<.001		0.003
Low	628	(85%)	75%		67%		60%		57%
Medium	930	(13%)	79%		71%		65%		59%
High	1710	(23%)	77%		69%		64%		58%
Unknown	4133	(56%)	75%		66%		59%		54%
Have a part-time joba				<.001		0.942		0.442		0.164
No	6238	(84%)	77%		67%		60%		55%
Yes	1112	(15%)	70%		68%		62%		58%
Academic self-evaluationb				<.001		<.001		<.001		<.001
Poor/very poor	205	(3%)	68%		57%		47%		47%
Average	1499	(20%)	71%		63%		54%		51%
Good	3583	(48%)	76%		68%		62%		55%
Excellent	1841	(25%)	82%		73%		65%		62%
Unknown	273	(4%)	73%		61%		55%		52%
Teachers care about mea				<.001		<.001		<.001		<.001
No	1269	(17%)	60%		58%		49%		46%
Yes	6027	(81%)	80%		70%		63%		58%
Ever felt bullied during this school yeara				<.001		<.001		0.001		<.001
No	3661	(49%)	74%		66%		59%		53%
Yes	3599	(49%)	78%		69%		63%		59%
Student council participationa				0.006		0.230		0.567		0.069
No	6325	(85%)	76%		67%		61%		55%
Yes	1025	(14%)	80%		69%		61%		58%
School sports participationa				0.840		0.006		0.183		0.001
No	4498	(61%)	76%		66%		60%		54%
Yes	2852	(39%)	76%		69%		62%		58%
School choir or band participationa				<.001		<.001		<.001		<.001
No	5772	(78%)	74%		66%		59%		53%
Yes	1578	(21%)	83%		73%		68%		66%
After school club participationa				<.001		<.001		<.001		<.001
No	4182	(57%)	73%		64%		56%		51%
Yes	3168	(43%)	81%		73%		67%		62%
After school activity in community participationa				<.001		<.001		<.001		<.001
No	4751	(64%)	74%		64%		56%		51%
Yes	2599	(35%)	81%		73%		70%		65%

^aThe sub-category percentages do not include the students who were missing a response regarding that sub-category characteristic, which was less than 2% of students across these factors.

^bThe ‘unknown’ sub-category represents students who were not asked about this factor in earlier versions of the survey or who did not provide a response in later versions of the survey.

Note: For all instances where p<.001 for the comparison of the prevalence of support for reporting among the specific subcategories of the variable for the particular adult category, there is a statistically significant difference at p< .05 based on the Bonferroni adjustment to correct for an inflated Type I error due to the large number of significance tests in the table.

Norm Misperception

On average across the cohorts where support for reporting was identified as normative to the specific categories of adults (that is, where more than 50% supported reporting), 29% to 42% of students erroneously believed that most peers would not support reporting to a specific category of adults despite support being normative (Table 3). Additionally, the average prevalence of misperceived norms still ranged from 28% to 39% amidst the four categories when using the more conservative determination of actual support for reporting being normative within a cohort in the norm misperception calculation. Again using data from all the cohorts where actual respondent support was normative, the average prevalence of norm misperception was greater among cohorts with students in years 7 or higher (ranging from 45% to 57% depending on the category of adults to whom a student supported reporting). Finally, even in cohorts where support was highly normative (>75% of students supported reporting), about one-quarter of students misperceived these positive norms.

Table 3.

Distribution of misperceiving the peer injunctive norm in cohorts where the actual injunctive norm is to support reporting a weapon to adults.

	Mean prevalence (and s.d.) of erroneously perceiving that most peer students did not support reporting to a specific category of adults
Cohort type	Head teacher		Police officer		Teacher or counsellor		Parent or another adult relative
All	29%	(13 p.p.)	38%	(11 p.p.)	41%	(10 p.p.)	42%	(9 p.p.)
Primary level	24%	(8 p.p.)	36%	(9 p.p.)	39%	(9 p.p.)	41%	(8 p.p.)
Upper level	49%	(7 p.p.)	49%	(9 p.p.)	45%	(7 p.p.)	57%	n/a
Prevalence of support for reporting was 76–100% (i.e., support highly normative)	22%	(7 p.p.)	26%	(4 p.p.)	25%	(3 p.p.)	25%	(3 p.p.)
Prevalence of support for reporting was 51–75% (i.e., support moderately normative)	45%	(9 p.p.)	42%	(10 p.p.)	43%	(9 p.p.)	44%	(7 p.p.)

In contexts where support for reporting to a specific adult category was identified as normative, 34% to 44% of students across the overall sample erroneously perceived that most students in their cohort did not think that students should tell a specific category of adults about seeing a knife or a gun at school (Table 4). However, the prevalence of norm misperception varied across student characteristics. For example, a greater percentage of students who did not participate in student group activities, as well as a greater percentage of students who felt that teachers did not care, misperceived norms as compared to students who were involved or felt that teachers cared, respectively. Nevertheless, at least a third of students across various student characteristics underestimated the commonness of peer support for reporting.

Table 4.

Prevalence of misperceiving the peer injunctive norm that supports telling a specific category of adults about seeing a weapon at school by student characteristics.

	Percent of students who misperceived their school cohort’s norm in support of reporting a weapon to a specific category of adults
Student characteristics	Head teacher	p-value	Police officer	p-value	Teacher or counsellor	p-value	Parent or another adult relative	p-value
All	34%		41%		44%		43%
Gender		.311		.571		.371		.104
Female	34%		41%		45%		42%
Male	35%		40%		44%		44%
Year in school		<.001		<.001		<.001		<.001
Year 4	18%		37%		32%		40%
Year 5	21%		35%		38%		41%
Year 6	20%		28%		35%		39%
Year 7	39%		41%		49%		49%
Year 8	43%		46%		53%		56%
Year 9	49%		49%		55%		55%
Year 10	53%		53%		53%		58%
Year 11	53%		56%		58%		-
Race/ethnicity		<.001		<.001		<.001		.008
White	28%		35%		40%		40%
Black	44%		49%		53%		46%
Asian	37%		43%		43%		42%
Other	38%		43%		50%		44%
Mixed	38%		41%		45%		42%
Unknown	32%		42%		46%		49%
Parent education		.546		.020		<.001		.699
Low	33%		41%		42%		45%
Medium	32%		37%		42%		42%
High	34%		39%		41%		42%
Unknown	35%		42%		47%		43%
Have a part-time job		.004		.491		.322		.154
No	34%		41%		44%		43%
Yes	38%		40%		46%		40%
Academic self-evaluation		<.001		<.001		<.001		<.001
Poor/very poor	46%		48%		58%		59%
Average	39%		45%		52%		46%
Good	34%		41%		43%		43%
Excellent	29%		36%		39%		39%
Unknown	33%		41%		41%		39%
Teachers care about me		<.001		<.001		<.001		<.001
No	50%		51%		57%		53%
Yes	31%		38%		41%		41%
Ever felt bullied during this school year		.182		.760		.710		.749
No	35%		41%		44%		43%
Yes	33%		40%		45%		42%
Student council participation		.013		.154		.080		.734
No	35%		41%		45%		43%
Yes	31%		39%		42%		42%
School sports participation		.946		.068		.867		<.001
No	34%		42%		45%		45%
Yes	34%		39%		44%		39%
School choir or band participation		<.001		<.001		<.001		<.001
No	36%		42%		46%		45%
Yes	27%		35%		38%		37%
School club participation		<.001		<.001		<.001		<.001
No	38%		45%		50%		48%
Yes	30%		35%		38%		38%
After school activity in community participation		<.001		<.001		<.001		<.001
No	37%		44%		49%		47%
Yes	29%		34%		37%		36%

Notes: This analysis did not include the school cohorts where the actual norm did not support reporting to a police officer (1 school cohort), teachers or counsellor (6 school cohorts) or parent or another adult relative (8 school cohorts). For all instances where p <.001 for the comparison of the prevalence of support among the specific subcategories of the variable for the particular adult category, there is a statistically significant difference at p< .05 based on the Bonferroni adjustment to correct for an inflated Type I error due to the large number of significance tests in the table.

Factors Associated with Personal Support for Reporting

Students were two to three times more likely to personally support reporting a weapon to an adult when they perceived support for reporting to be normative as compared to students who did not perceive support for reporting to be normative (Figure 1). This dramatic difference was consistent and statistically significant across categories of potential adults to whom a student could report seeing a weapon at school such as a knife or a gun. According to estimates from the initial multilevel multivariate logistic regression a similar pattern of results was found. In addition, estimates indicated that for every percentage point increase in the prevalence of actual support in one’s cohort for reporting to a teacher or counsellor and parent or another adult relative, the likelihood of supporting reporting went up 2% and 1% respectively. There was no association between the actual prevalence of support and the outcome when in reference to support for reporting a weapon to a head teacher or a police officer. Analyses using the ‘minimum prevalence of support’ to calculate the actual norm level indicated similar associations with personal attitude for each outcome.

Fig. 1.

The prevalence of students who personally supported student reporting of weapons to specific categories of adults by whether they perceived the injunctive norm in one’s school cohort to support reporting.

The main regression analyses showed that the attitude students perceived to be normative in their school was by far the strongest predictor of a student’s personal support for reporting (Table 5). Students who believed that support for telling an adult was the injunctive norm in their school were 14 to 26 times more likely (depending on the category of adult) to personally believe students should tell an adult about seeing a weapon at school. In contrast, the strength of the association between actual injunctive norm levels and personal attitude was much smaller. For example, a student in a cohort with a highly normative level of support for telling a teacher or counsellor about a weapon was 2.6 times (95% CI 1.6, 4.1) more likely to personally believe students should tell them as compared to students in cohorts where the level of support among peers was not normative (50% or less prevalence). Students in cohorts where support for reporting to a teacher or counsellor was moderately normative were 1.3 times more likely to support reporting (95% CI 1.1, 1.6). The associations between the actual injunctive norm levels and personal support for reporting to other categories of adults were of slightly less magnitude and not statistically significant.

Table 5.

Mutually adjusted odds-ratios (logistic regression) for perceived and actual injunctive norms predicting personally believing that students should tell specific categories of adults about seeing a weapon at school.

	Supported reporting a weapon at school to a specific category of adults
	Head teacher		Police officer		Teacher or counsellor		Parent or another adult relative
	AOR	(95% CI)	AOR	(95% CI)	AOR	(95% CI)	AOR	(95% CI)
Accurately perceived that most cohort peers supported reporting a weapon to adult (vs. misperceived norm as not in support of reporting)	13.78***	(11.9, 15.92)	25.88***	(22.40, 29.90)	18.87***	(16.54, 21.53)	22.97***	(19.99, 26.38)
Highly normative actual support for reporting (75% prevalence or more) among cohort peers (vs. actual support for reporting not normative with only 50% or less prevalence)	1.20	(0.91, 1.59)	1.27	(0.98, 1.64)	2.56***	(1.59, 4.13)	1.51	(0.84, 2.71)
Moderately normative actual support for reporting (51–75% prevalence) among cohort peers (vs. actual support for reporting not normative with only 50% or less prevalence)	--a	--a	--b	--b	1.30**	(1.07, 1.58)	0.97	(0.77, 1.22)
Agreed that teachers care about student (vs. not)	1.69***	(1.43, 2.00)	1.16	(0.97, 1.38)	1.28**	(1.09, 1.51)	1.14	(0.96, 1.35)
Felt bullied at least once during school year (vs. not)	1.15*	(1.00, 1.32)	1.30***	(1.14, 1.49)	1.30***	(1.15, 1.48)	1.17**	(1.04, 1.34)
Participated in after school club (vs. not)	1.25**	(1.08, 1.45)	1.17*	(1.02, 1.35)	1.17*	(1.02, 1.34)	1.15*	(1.01, 1.32)
Participated in school choir or band (vs. not)	1.26*	(1.05, 1.52)	1.14	(0.96, 1.37)	1.25**	(1.06, 1.47)	1.43***	(1.22, 1.68)
Participated in school council (vs. not)	1.05	(0.86, 1.29)	1.02	(0.83, 1.24)	0.95	(0.79, 1.14)	1.04	(0.86, 1.25)
Participated in school sports (vs. not)	0.96	(0.83, 1.11)	1.10	(0.95, 1.27)	0.97	(0.85, 1.11)	1.07	(0.93, 1.22)
Participated in after school activity in community (vs. not)	1.06	(0.91, 1,24)	1.06	(0.91, 1.24)	1.37***	(1.20, 1.58)	1.26**	(1.10, 1.45)
Having a part-time job (vs. not)	0.77**	(0.64, 0.93)	0.96	(0.79, 1.17)	1.05	(0.88, 1.25)	1.16	(0.97, 1.39)
Year in school	0.87 ***	(0.82, 0.92)	1.00	(0.95, 1.04)	1.06*	(1.01, 1.10)	0.96	(0.92, 1.01)
Female (vs. male)	1.16*	(1.00, 1.35)	0.85*	(0.74, 0.98)	1.13	(0.99, 1.29)	1.03	(0.90, 1.18)
Parental education (vs. low)
Medium	1.56**	(1.16, 2.10)	1.20	(0.90, 1.61)	1.36*	(1.03, 1.79)	1.22	(0.92, 1.60)
High	1.32*	(1.01, 1.74)	1.06	(0.81, 1.38)	1.24	(0.96, 1.59)	1.01	(0.78, 1.30)
Unknown	1.22	(0.94, 1.58)	1.02	(0.79, 1.31)	1.21	(0.96, 1.55)	0.99	(0.78, 1.27)
Race/Ethnicity (vs. Black)
White	1.49***	(1.23, 1.81)	1.22*	(1.01, 1.46)	1.28**	(1.08, 1.53)	1.14	(0.95, 1.36)
Asian	1.38**	(1.09, 1.75)	1.14	(0.90, 1.44)	1.19	(0.95, 1.48)	1.08	(0.86, 1.35)
Mixed	0.88	(0.69, 1.12)	1.05	(0.82, 1.34)	0.91	(0.72, 1.14)	1.05	(0.83, 1.32)
Other	1.00	(0.65, 1.52)	1.19	(0.77, 1.83)	1.08	(0.72, 1.62)	1.00	(0.66, 1.51)
Unknown	1.30	(0.98, 1.71)	1.08	(0.83, 1.42)	0.94	(0.73, 1.21)	1.23	(0.95, 1.59)
GPA (vs. poorly/very poorly)
Average	0.80	(0.54, 1.19)	1.28	(0.85, 1.94)	1.10	(0.75, 1.61)	0.82	(0.56, 1.20)
Good	0.94	(0.64, 1.38)	1.59*	(1.07, 2.38)	1.36	(0.94, 1.97)	0.95	(0.66, 1.37)
Excellent	1.13	(0.76, 1.68)	1.78**	(1.17, 2.70)	1.52*	(1.04, 2.24)	1.15	(0.78, 1.69)
Unknown	0.77	(0.45, 1.32)	1.27	(0.73, 2.20)	1.42	(0.85, 2.39)	0.90	(0.53, 1.52)

^aThere were no cohorts in this category.

^bThere was only one cohort in this category so it was not included in this model.

Note: Data were fitted to a multilevel logistic regression model accounting for clustering of support within school cohorts for reporting a weapon to a specific category of adults and also adjusting for the year that the survey was taken.

^*p < .05;

^**p < .01;

^***p < .001 (two-tailed tests).

Finally, estimates from the main models indicated that students who had felt bullied at school and who participated in school clubs were 1.15 to 1.30 times more likely to say students should report seeing a weapon at school to each of the categories of adults. In contrast, students who participated in school sports or who participated in school council were no more or less likely to support reporting compared to students who did not participate in these activities. The statistical relevance of other explanatory factors varied depending on the specific category of adults to whom students supported reporting. For example, students in higher years in school were less likely to support reporting a weapon to a head teacher than students in lower years, but this was not found for other adult categories.

DISCUSSION

Little is known about whether students misperceive peer injunctive norms about support for reporting weapons at school to an adult across a variety of student groups and contexts, and if so, whether perceived normative support is associated with personal support for reporting when adjusting for actual levels of support at school. Given decades of research on social conformity among humans and peer influence among adolescents, empirically determining the extent of peer norm misperceptions about being an active bystander at school is critical for understanding how to increase reporting potentially violent threats among youth in schools. Based on social norms theory, this study assessed personal attitudes (whether a student supported reporting a weapon at school to a variety of adults including head teacher, teacher or counsellor, police officer, and parent or another adult relative) and perceived injunctive norms (whether a student perceived most peers to support reporting) among 7,401 students in years 4 to 11 across 45 cohorts in 33 schools in a Greater London borough. Actual injunctive norms were classified as supportive of reporting if the prevalence of support was present among more than 50% of the cohort. Peer norm misperception was assessed by comparing actual norms with perceived norms about one’s cohort. The prevalence of support for reporting as well as misperception about support for reporting norms were displayed across student characteristics, activities, and contexts to indicate whether support for being an active bystander was widely reported across student populations while also being misperceived by many students as atypical among peers. In addition, the association between a perceived injunctive norm and personal attitude was assessed while adjusting for other factors and the actual level of support within the cohort to understand the extent to which perception of a norm is a separate construct from the actual norm and how crucial perceived norms might be in determining personal support. Misperceived norms about support for reporting across a range of students and contexts and evidence of a strong association between perception and personal attitude would indicate an opportunity for a population-wide intervention to correct misperceptions and potentially increase reporting. Such findings would be important for violence prevention work as many factors associated with support for being an active bystander (including the other factors assessed in this study) are ostensibly not as easy to change (if at all).

This study’s first hypothesis was confirmed as 34% to 44% of students erroneously believed that most other students in their school would not say that students should tell a head teacher, teacher or counsellor, police officer, or parent or another adult relative about seeing a weapon such as a knife or a gun at school, despite actual injunctive norms that supported reporting weapons to adults (among most primary and upper level cohorts). Norm misperceptions were substantially present across a range of students and contexts. In addition, the data suggested that as adolescents grow older and move into a more peer intensive time periods in their lives, they simultaneously become less accurate in knowing actual peer norms about support for being an active bystander when weapons are brought to school. Supporting the second hypothesis, results indicated that what a student perceived to be the injunctive norm in their cohort about support for reporting a weapon to a specific category of adults was a far stronger predictor of personally believing that students should report a weapon than was the actual injunctive norm. Thus, if what is perceived as real is real in its consequences (Thomas and Thomas 1928), then the pervasive misperception of peer aversion to telling an adult about seeing a weapon at school is problematic. Norm misperceptions may diminish personal support for being an active bystander and ultimately act as a barrier to prosocial engagement and violence prevention by encouraging non-reporting.

Results are similar to findings from a study of middle and high school youth in the United States where a substantial amount of students underestimated peer support for reporting weapons to adults and that perceived injunctive norms were associated with personal attitudes toward reporting (J.M. Perkins et al. 2017a). In addition, a study of youth in five middle schools in New Jersey suggested that increases in accuracy of perceived norms about reporting bullying harassment to adults were associated with increases in personal support for reporting (H. W. Perkins et al. 2011). The current study adds to the literature by identifying the extent of norm misperceptions and their potential influence in a culture that has much less gun violence among school-aged youth, but greater challenges with knife-related violence (Allen and Audickas 2018; Vulliamy et al. 2018). In addition, this study is conducted in contexts of much greater ethnic diversity across 35 schools of greater London than is present in most of the US schools included in past studies related to this work. Finally, the prevalence of norm misperception is examined across students representing a large set of sociodemographic characteristics and extracurricular activities and the associations of both actual and perceived norms with personal attitudes are assessed simultaneously while also adjusting for many student factors. From an applied practice vantage point, understanding the role that (mis)perceived peer norms may play across the general student population in preventing youth from becoming active bystanders is a novel entry point for violence prevention work in the UK.

Potential Causes of Misperception

Norm misperceptions may likely reflect distorted images provided by media and video games that may make weapon carrying and violence in schools seem normal and acceptable and therefore not a threat to be reported. Institutional signals from a school or from government may also contribute to norm misperceptions (Tankard and Paluck 2016). If official communications about violence or prevention structures such as metal detectors and heightened security are designed on a fear-based response, then it seems as if individuals are not supporting or engaging in protective bystander behavior. For example, the substantial attention placed on knife stabbings in the UK or the attention on school shootings in the US may create perceptions that schools are dangerous places where threats are not reported despite the rarity of weapons-based violence overall (Cornell 2015). In addition, the media typically do not emphasize positive normative attitudes or behaviors as such information is not what is thought to make news and capture people’s attention. Moreover, advocacy of a positive attitude or behavior in the media may typically be discussed in the context of fear, which does not make the positive attitude or behavior seem normative. Thus, youth (and even adults) may think that weapon-related violence is more prevalent than it actually is in their community and therefore also think that not much can be done about it (i.e., not worth telling an adult or authority about a potential threat). For example, one study among youth in the UK found that they perceived gun and knife crimes to be increasing in their local areas despite decreases in such crimes (Minty, Hollingworth, Allen, and Holden 2010), and another study found that students felt as if violence was unpreventable (Sundaram 2016).

Second, peer communication may lead to misperceptions of social norms (H. W. Perkins 2003a; Lapinski and Rimal 2006). For example, students may disproportionately talk about saying nothing or about peers who said nothing before an event occurred despite knowing something. Students may therefore interpret staying silent as normal. Continuous talk itself about the event may normalize the event and make it seem like it is difficult to prevent violence, thus encouraging silence. In addition, if the communication that is present does not provide a complete picture of information about peers’ attitudes and behaviors across a variety of contexts, then such communication may lead to attribution error and further reduce conversation about support for telling authorities about threats. A recent study among young Germans found that both communication content and frequency were associated with the perception of social norms about risky driving behaviors (Geber, Baumann, and Klimmt 2017).

Third, family, teachers, or other influential role models, may use language that makes it seem like few students support reporting or that students are tolerant of peers who carry weapons. In addition, some youth may develop norm misperceptions about the student body based on what they perceive to be close friends’ beliefs or actions, which may represent an uncommon view. Within certain social network structures, this locally presented but globally uncommon outcome can contribute to a majority illusion where people in these networks overestimate the prevalence of the outcome across the larger network (Lerman, Yan, and Wu 2016).

Implications for Intervention

Given the importance of active bystanders in preventing violent incidents, the one-quarter to one-half of students in this study who did not personally support reporting suggests intervention is needed. The results of this study suggest an opportunity to take a social norms approach in designing a population-wide, school-based initiative to reduce perceived social barriers to student bystander engagement. Communicating that most of the student body believes that students should tell different types of adults about seeing a weapon at school may correct misperceptions among those students who were unaware of the most common attitudes about reporting held by most peers. If there is a causal link between norm perception and personal attitude, then these changes in perception could cause positive changes in students’ personal attitudes about telling an adult if they did not previously support reporting. This kind of intervention might then also lead to increased reporting as attitudes are important predictors of behavior. In addition, changes in norm perception could reinforce the attitudes of students who already supported reporting but who had misperceived their peers’ attitudes (i.e., who were previously carriers of the misperception). These students––perhaps feeling more empowered––might be more vocal in their attitudes, thus further supporting a climate that would openly reinforce positive attitudes and provide support for behavioral change at later time points among students whose attitudes did not initially change in response to the intervention.

Social norms interventions could focus on school-wide dissemination of true positive norms via media such as screen savers, posters, weekly news messages, websites, other online social media pages, or twitter. In addition, sending true SMS text-messages like “Most students in this school think students should tell a teacher if they see a weapon at school” might be effective, particularly if they are made aware that the claims are based on data collected in their own local peer population. Furthermore, school counsellors or nurses could provide personalized normative feedback to students when they have appointments for behavioral issues or other violent incidents, etc. Alternatively, interventions could provide peer leaders with accurate norm information and the tools to help spread this information and model peer support for reporting (Paluck and Shepherd 2012).These norms-based communications may help increase the impact of other violence prevention interventions, such as the implementation of a 24-hour hotline or bystander training workshops. Students may be more likely to make a call if they know that their peers would generally support them doing so or may more enthusiastically participate in bystander training knowing that others would support intervening and reporting a threat.

Future Research Directions

Results from this study present a foundation for further research on misperceiving peer norms about weapons-related violence and bystander attitudes and behavior. For example, studying perceived descriptive norms about actually reporting threats to adults and authorities would provide further evidence to build interventions correcting misperceptions (assuming perceived norms predict both behavior and attitudes). In addition, asking questions about specific scenarios, such as seeing an unknown student have a weapon versus a friend having one, might reveal a variety of norm misperceptions that could provide opportunities for intervention and correction. Also, asking questions about what students do not support and perceptions of what most other students do not support, such as “staying silent” might show that both ‘do and don’t norms’ are important to consider when designing anti-violence norms-based prevention material (Bergquist and Nilsson 2019). Finally, if norm misperceptions about common attitudes and behaviors among different peer reference groups are identified, then the positive norms existing among a variety of groups could be highlighted in an intervention and be self-reinforcing. Although the prevalence or extent of norm misperception about close social reference groups might be less than the norm misperception about more distal social reference groups, both types of norm misperception may provide intervention opportunities if perceived norms about multiple groups are all associated with personal attitudes and behavior.

Study Limitations

Interpretation of findings is subject to limitations. First, the data were self-reported. However, attitudes are not subject to recall error and there is little incentive to lie under the anonymous data collection design (or confidential design depending on the format). Second, there could be some small bias arising through response options that had an ‘or’ included (e.g. teacher or a counsellor) as students might have chosen the option (or not) due to one of the adult types presented in that option and not the other. Third, results should not be generalized to all UK students or to youth in other countries. However, the pervasiveness of misperceived norms across a wide range of students attending a diversity of schools is provided. Moreover, a high response rate from the entire population of the schools’ target years provides confidence in the representativeness of results for these schools. Thus, it is not likely that the main findings of this paper would change substantially based on a nationally representative sample with a somewhat different demographic distribution. Finally, directionality cannot be confirmed here as data were cross-sectional. Although research on other topics has found evidence that perceived norms impact personal changes in attitude and behavior (as cited previously), longitudinal and experimental data are needed to determine how perceived norms of support for telling adults about weapons at school influence personal attitudes, and, ultimately, reporting behavior.

Conclusion

As youth may be aware of weapons brought to school by peers, building support for reporting these threats to adults is a crucial part of violence prevention. This study found that many youth underestimate support among peers for telling various adults about seeing a weapon such as knife or a gun at school. Correcting these norm misperceptions is important because accurately thinking that most peers support reporting is strongly associated with personal support for reporting. Researchers and practitioners interested in weapons-based violence prevention among school-aged youth should assess whether misperceived injunctive norms about being a prosocial bystander within the school setting exist. If so, then reducing norm misperceptions may help to break down an erroneously perceived norm endorsing a code of silence. Students may then be more likely to positively respond to school policies encouraging them to report if they know someone has brought a weapon to school. School administrators and policymakers should explore methods to promote awareness of positive actual norms among the student body. Work to reduce these misperceived norms should begin in the early years of adolescence and persist through the later years in secondary schools. The increase in misperceptions across years might naively appear to be inevitable as a developmental outcome of the intensification of peer interaction. The greater prevalence of norm misperceptions in later adolescence might also suggest, however, that a concentration of effort to reduce misperceptions among older youth is worthwhile. The social forces contributing to these misperceptions begin to distort norm perceptions at young ages and then continue to exacerbate the misperceptions in later adolescence. Thus, any reduction in misperceived norms achieved among students at any age in school may help arrest or slow the growth of increasingly problematic misperceptions later and make interventions to correct norm misperceptions among older students less challenging and more effective.